watcher-hermes/hermes-skills/ai-sre-plan
Claude Code and agent skill by watcher-hermes/hermes-skills
- community
Checking that the SKILL.md file is still in the repository.
Design an AI SRE rollout for a team — multi-agent triage architecture, structured runbooks, adversarial evaluation, narrow auto-remediation, and predictive-detection posture.
Install
npx vskill@latest install watcher-hermes/hermes-skills/ai-sre-planSKILL.md
The same SKILL.md was published earlier as watcher-hermes/hermes-skills/ai-sre-plan.
Given team size, incident volume, observability maturity, and risk tolerance, produce an AI SRE plan.
Produce:
- Architecture. Multi-agent: supervisor + log agent + metric agent + runbook agent + human gate. Match specialized agents to existing data sources (Datadog, Grafana, Loki, Confluence).
- Runbook transformation. Move from unstructured Confluence to structured markdown with symptom / hypothesis / verify / act sections. Version in git.
- Product choice. Datadog Bits AI, Azure SRE Agent, NeuBird Hawkeye, Incident.io Autopilot, or DIY.
- Auto-remediation scope. Narrow safe set (restart pod, revert deploy, scale within bounds). Explicit deny list (topology, code, IAM, database). Policy as code.
- Adversarial evaluation. Specify two-model agreement gate for auto-remediation. Disagreement escalates.
- Predictive-detection posture. If considering (MIT 89% result), name the actuation policy — pager, pre-drain, auto-scale — otherwise it's just a dashboard.
Hard rejects:
- Auto-remediation without human gate on broad changes. Refuse — name the safe set explicitly.
- Unstructured runbooks as the knowledge base. Refuse — require structured, versioned markdown.
- "Set it and forget it" framing. Refuse — explicitly scope what is and isn't autonomous.
Refusal rules:
- If incident volume is <10/month, refuse full AI SRE rollout — cost exceeds benefit. Recommend structured runbooks only.
- If team observability is immature (logs unsearchable, metrics sparse), refuse — AI SRE amplifies bad data.
- If the team proposes "predictive detection → auto-remediation" as first feature, refuse — walk through the actuation-policy question first.
Output: a one-page plan with architecture, runbook plan, product choice, auto-remediation scope, adversarial gate, predictive posture. End with a 12-week rollout schedule: weeks 1-4 structured runbooks, 5-8 triage agent, 9-12 narrow auto-remediation.
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/ai-sre-plan/SKILL.md
Popularity
- 7-day trend score
- 5
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/watcher-hermes/hermes-skills/ai-sre-plan)